Parameterized Two-Player Nash Equilibrium
|
|
- Kristopher Pierce
- 5 years ago
- Views:
Transcription
1 Parameterized Two-Player Nash Equilibrium Danny Hermelin, Chien-Chung Huang, Stefan Kratsch, and Magnus Wahlström Max-Planck-Institute for Informatics, Saarbrücken, Germany Abstract. We study the problem of computing Nash equilibria in a two-player normal form game from the perspective of parameterized complexity. Recent results proved hardness for a number of variants, when parameterized by the support size. We complement those results, by identifying three cases in which the problem becomes fixed-parameter tractable. These cases occur in the previously studied settings of sparse games and unbalanced games as well as in the newly considered case of locally bounded treewidth games that generalizes both these two cases. 1 Introduction Algorithmic game theory is a quite recent yet rapidly developing discipline that lies at the intersection of computer science and game theory. The emergence of the internet has given rise to numerous applications in this area such as online auctions, online advertising, and search engine page ranking, where humans and computers interact with each other as selfish agents negotiating to maximize their own payoff utilities. The amount of research spent in attempting to devise computational models and algorithms for studying these types of interactions has been overwhelming in recent years; unsurprisingly perhaps, when one considers the economical rewards available in this venture. The central problem in algorithmic game theory is that of computing a Nash equilibrium, a set of strategies for each player in a given game, where no player can gain by changing his strategy when all other players strategies remain fixed. This problem is so important because Nash equilibria provide a good way to predict the outcomes of many of the scenarios described above, and other scenarios as well. Furthermore, Nash s Theorem states that for any finite game a mixed Nash equilibrium always exists. However, for this concept to be meaningful for predicting behaviors of rational agents which are in many cases computers, a natural prerequisite is for it to be computable. This led researchers such as Papadimitriou to dub the problem of computing Nash equilibria as one of the most important complexity problems of our time [27]. The initial breakthrough in determining the complexity of computing Nash equilibria was made by Daskalakis, Goldberg, and Papadimitriou [12, 23]. These two papers introduced a reduction technique which was used by the authors for showing that computing a Nash equilibrium in a four player game is PPADcomplete. Shortly afterwards, this hardness result was simultaneously extended to three player games by Daskalakis and Papadimitriou [16], and by Chen and
2 2 Deng [6]. The case of two player (bimatrix) games was finally cracked a year later by Chen and Deng [7], who proved it to be PPAD-complete. This implied that the existence of a polynomial-time algorithm for the core case of bimatrix games is unlikely. Since the result of Chen and Deng [7], the focus on computing Nash equilibria in bimatrix games was directed either towards finding approximate Nash equilibria [4, 8 10, 13, 14, 25], or towards finding special cases where exact equilibria can be computed in polynomial time [2, 9, 11, 24, 25]. Nevertheless, for general bimatrix games the best known algorithm for computing either approximate or exact equilibria essentially tries all possibilities for the support of both players (the set of strategies played with non-zero probability).once the support of both players is known, one can compute a Nash equilibrium by solving a linear-program. Theorem 1 ([26]). A Nash equilibrium in a bimatrix game, where the support sizes are bounded by k, can be computed in n O(k) time. Due to the central role that the algorithm of Theorem 1 plays in computing exact and approximate Nash equilibria, it is natural to ask whether one can improve on its running-time substantially. In particular, can we remove the dependency on the support size from the exponent? The standard framework for answering such questions is that of parameterized complexity theory [17, 20]. Estivill-Castro and Parsa initiated the study of computing Nash equilibria in this context [19]. They showed that when the support size is taken as a parameter, the problem is W[2]-hard even in certain restricted settings. The implication of their result is a negative answer to the above question. In particular, combining their reduction with the results of Chen et al. [5] gives a sharp contrast to Theorem 1 above. Theorem 2 ([19]). Unless FPT=W[1], there is no n o(k) time algorithm for computing a Nash equilibrium with support size at most k in a bimatrix game. The consequence of Theorem 2 above is devastating in the sense that for large enough games that have equilibriums with reasonably small supports, the task of computing equilibria already becomes infeasible. The main motivation of this paper is to find scenarios where one can circumvent this. Our goal is thus to identify natural parameters which govern the complexity of computing Nash equilibria, and which can help in devising feasible algorithms. We believe that this direction can prove to be fruitful in the quest for understanding the computational limitations of this fundamental problem. Indeed, prior to our work, Kalyanaraman and Umans [24] provided a fixed-parameter algorithm for finding equilibrium in bimatrix games whose matrices have small rank (and some additional constraints). Our techniques are based on considering a natural graph-theoretic representation of bimatrix games. This is done by taking the union of the underlying boolean matrix of the two given payoff matrices, and considering this matrix as the biadjacency matrix of a bipartite graph. A similar approach was taken by [11], and in particular by [2] who considered games that have an underlying
3 3 planar graph structure. Our work complements both these results as will be explained further on, and further exemplifies the strength of a graph-theoretical approach when computing Nash equilibria in bimatrix games. A natural class of games that has a convenient interpretation in the graphtheoretic context is the class of l-sparse games [9, 11, 15]. Here each column and row in both payoff matrices of the game have at most l non-zero entries. An initial tempting approach in these types of games would be to try to devise a parameterized algorithm with l taken as a single parameter. However, Chen, Deng, and Teng [9] showed that unless PPAD = P, there is no algorithm for computing an ε-approximate equilibrium for a 10-sparse game in time polynomial both in ε and n. Thus, such an FPT algorithm cannot exist unless PPAD is in P. We complement this result by showing that if l is taken as a parameter, and the size of the supports is taken as an additional parameter, then computing Nash equilibrium is fixed-parameter tractable. Theorem 3. A Nash equilibrium in a l-sparse bimatrix game, where the support sizes is bounded by k, can be computed in l O(kl) n O(1) time. Note that the above result also complements the polynomial time algorithms given in [9, 11] for 2-sparse games. While in these algorithms there was no assumption made on the size of support of the equilibrium to be found, both algorithms could handle only win-lose games [1, 10], games with boolean payoff matrices. Theorem 3, on the other hand, holds for arbitrary payoffs. Our second result is concerned with k-unbalanced games, games where the row player has a small set of k strategies [24, 25]. Lipton, Markakis, and Mehta [25] observed that in such games there is always an equilibrium where the row player plays a strategy with support size at most k + 1. Thus, by applying Theorem 1 one can find a Nash equilibrium in n O(k) time for these types of games. Can this result be improved to an algorithm running in f(k) n O(1) time? We give a partial answer to this question, by showing that if the number l of different payoffs of the row player is taken as an additional parameter, the problem indeed becomes fixed-parameter tractable. Theorem 4. A Nash equilibrium in a k-unbalanced bimatrix game, where the row player has l different payoff values, can be computed in l O(k2) n O(1) time. In our last result, examining the borderline of FPT cases, we consider a structural property that simultaneously extends both the previous cases (albeit only in the case of a bounded number of different payoffs). We show that for bimatrix games whose corresponding graph has locally bounded treewidth, and where the payoff matrices have at most l different values, we can compute a Nash equilibrium of support size at most k in time f(k, l) n O(1). In addition to the above cases of sparse and unbalanced games, this also includes many other cases including games where the underlying graph structure is planar, as considered by [2]. However, as this class is quite general, the running-time dependency on both parameters is worse.
4 4 Theorem 5. A Nash equilibrium in a locally bounded treewidth game, where the support sizes are bounded by k, and the payoff matrices have at most l different values, can be computed computed in f(k, l) n O(1) time for some computable function f(). The paper is organized as follows: We begin with some preliminaries in Section 2. In Section 3 we consider l-sparse games and prove Theorem 3. Section 4 addresses locally bounded treewidth games and proves Theorem 5. Finally, in Section 5 we prove Theorem 4 regarding k-unbalanced games. 2 Preliminaries Let G := (A, B) be a bimatrix game, where A, B Q n n are the payoff matrices of the row and the column players respectively. The row (column) player has a strategy space consisting of the rows (columns) [n] := {1,..., n}. (For ease of notation, except in unbalanced games, we assume that both players have the same number of strategies; different numbers of strategies do not affect any of our results.) The row (column) player chooses a strategy profile x (resp. y), which is a probability distribution over his strategy space. That is, x i, y j 0 for all i, j [n], and furthermore n i=1 x i = 1 and n j=1 y j = 1. The expected outcomes of the game for the row and the column players are x T Ay and x T By respectively. The players are rational, always aiming for maximizing their expected payoffs. They have reached a Nash equilibrium if the current strategies x and y are such that neither player has a deviating strategy ˆx or ŷ such that ˆx T Ay > x T Ay or x T Bŷ > x T By. In other words, if neither of them can improve his payoff independently of the other. The following proposition gives an equivalent condition for a pair of strategies to be an equilibrium. Lemma 1. ([26, Chapter 3]) The pair of strategy vectors (x, y) is a Nash equilibrium for the bimatrix game (A, B) if and only if (i) x s > 0 (Ay) s (Ay) j for all j s; (ii) y s > 0 (x T B) s (x T B) j for all j s. The support of a strategy vector x is defined as the set S(x) = {i : x i > 0}. Note that the above proposition implies that if (x, y) is a Nash equilibrium, in the column vector Ay, the entries in S(x) are equivalent and no less than all other entries not in S(x); symmetrically, in the row vector x T B, the entries in S(y) are equivalent and no less than other entries not in S(y). It is known that, given possible supports I, J [n] it can be efficiently decided whether there is a matching Nash equilibrium, and the corresponding strategy vectors can be computed via linear programming. The following graph associated with a bimatrix game is useful for presenting our algorithms in Sections 3 and 4.
5 5 Definition 1. Let G = (A, B) be a bimatrix game with A, B Q n n. The undirected bipartite graph G := G(G) associated with G is defined to be the bipartite graph with vertex classes V r := {r 1,..., r n } and V c := {c 1,..., c n }, referred to as row resp. column vertices, where r i V r and c j V c are adjacent in G iff A i,j 0 or B i,j 0. As a last bit of notation: For I, J [n], and any n n matrix A, we use A I,J to denote the submatrix composed of rows in I and columns in J. We also use A I, as a shorthand for A I,[n]. Thus, A i, means the i th row of A. 3 Sparse Games In this section we present the proof for Theorem 3. Throughout the section we let G := (A, B) denote our given bimatrix game, where A and B are rational value matrices with at most l non-zero entries per row or column. We will present an algorithm for finding an Nash equilibrium where the support sizes of both players are at most k (and k is taken as a parameter). The high-level strategy is to show that it suffices to search for equilibria that induce one or two connected components in the associated graph G = G(G). This permits us to find candidate support sets by enumerating subgraphs of G (on one or two components). Central to this strategy is the notion of minimal equilibria: Definition 2. A Nash equilibrium (x, y) is minimal if for any Nash equilibrium (x, y ) with S(x ) S(x) and S(y ) S(y), we have S(x ) = S(x) and S(y ) = S(y). Our algorithm iterates through all possible support sizes k 1, k 2 k in increasing order to determine whether there exists an equilibrium (x, y) with S(x) = k 1 and S(y) = k 2. To avoid cumbersome notation, we will assume that k 1 = k 2 = k (extending this to general case will be immediate). Thus at a given iteration, the algorithm can assume that no equilibrium exists with smaller supports, which means it can restrict its search to minimal equilibriums. This fact will prove crucial later on. Furthermore, observe that we can assume n > lk since otherwise obtaining the running time required by Theorem 3 is trivial using Theorem 1. Therefore, since our game is l-sparse, our algorithm only needs to search for equilibriums where both players receive non-negative payoffs. Lemma 2. If G = (A, B) is an l-sparse game, where A, B Q n n and n > lk, then in any Nash equilibrium with support at most k k, both players receive non-negative payoffs. Proof. Suppose (x, y) is an equilibrium with S(x) k and S(y) k, where the row player has a negative payoff. Since the A is l-sparse, and n > lk, there exists an all zero row i 0 [n] in A,S(y). Thus, letting the row player choose the strategy vector ˆx with ˆx i0 = 1 and ˆx i = 0 for all i i 0, we get that ˆx T Ay = 0 > x T Ay, contracting the fact that (x, y) is an equilibrium. The argument for the column player is symmetric.
6 6 For an equilibrium (x, y), let the extended support of x be the rows S(x) N(S(y)), and similarly for y, where the neighborhood N(I) is taken over the graph G := G(G) of the game. Note that any row not in the extended support of x would have payoff constantly zero given the current strategy of y, and thus is not important for the existence of an equilibrium. We will show that for a minimal equilibrium (x, y), the extended supports of x and y induce a subgraph of G which has at most two connected components. This will be done in two steps: The first is the special case where A S(x),S(y) = B S(x),S(y) = 0, while the second corresponds to the remaining cases. Lemma 3. If (x, y) is a minimal Nash equilibrium for a game (A, B) with A S(x),S(y) = B S(x),S(y) = 0, then the subgraphs induced by N[S(x)] and N[S(y)] in the graph associated with the game are both connected. Proof. Let H := G[N[S(x)]] be the subgraph of G induced by N[S(x)], and aiming towards a contradiction, suppose that H is disconnected. Let C be a connected component in H, and write p := i V r (C) x i to denote the probability that a row strategy in C is played according to x. Now define a new row strategy by setting ˆx i = x i /p if i V r (C), and ˆx i = 0 otherwise. We argue that (ˆx, y) is a Nash equilibrium, contradicting the fact that (x, y) is minimal. Obviously, the expected payoff in (ˆx, y) is zero for both players, as A S(ˆx),S(y) = B S(ˆx),S(y) = 0. Furthermore, there is no row i such that (Ay) i > 0, since the strategy y is unchanged and the original strategy pair (x, y) is an equilibrium. Now assume that there is a column j such that (ˆx T B) j > 0. Then B i,j 0 for some i S(ˆx), and by the connectivity assumption we must have B i,j = 0 for all i S(x) \ S(ˆx), but then (x T B) j = p(ˆx T B) j > 0, which contradicts the assumption that (x, y) is a Nash equilibrium. By Lemma 1, it follows that (ˆx, y) is a Nash equilibrium. Lemma 4. If (x, y) is a minimal Nash equilibrium for a game (A, B) with either A S(x),S(y) 0 or B S(x),S(y) 0, and with a non-negative payoff for both players, then the subgraph induced by N[S(x) S(y)] is connected. Proof. Let H be the subgraph induced by N[S(x) S(y)] and suppose that H is disconnected. We will derive a contradiction by showing that (x, y) is not minimal. Let C be a connected component in H intersecting both S(x) and S(y), and let Vˆx := V r (C) S(x) and Vŷ := V c (C) S(y) denote the set of row and column strategies in C, respectively. We define a new pair of strategy profiles (ˆx, ŷ) where S(ˆx) = Vˆx and S(ŷ) = Vŷ, by normalizing (x, y) onto Vˆx and Vŷ. That is, we let p := x i Vˆx i, and set ˆx i = x i /p if i Vˆx, and ˆx i = 0 otherwise. Similarly, we let q := y j Vŷ i, and set ŷ accordingly. As either S(ˆx) S(x) or S(ŷ) S(y), to prove the lemma it suffices to argue that (ˆx, ŷ) is an equilibrium. Consider a row strategy i [n]. We claim that { (Ay) i /q if A i,vŷ 0, (Aŷ) i = (1) 0 otherwise.
7 The second case is clear. For the first case, assume that A i,vŷ 0. Now, if A i,j 0 for j S(y) \ Vŷ, then there would be an edge in H from the vertex corresponding to row i, which is in C, to the vertex corresponding to column j, which is not in C, contradicting that C is a connected component. Thus (Ay) i = A i,j y j = A i,j (qŷ j ) + A i,j y j = q(aŷ) i + 0 j S(y) j Vŷ j S(y)\Vŷ and the claim follows. Now let s S(ˆx), and consider some arbitrary row strategy i [n]. Assume by way of contradiction that (Aŷ) s < (Aŷ) i. It is clear that A i,vŷ 0, thus (Aŷ) i = (Ay) i /q; we consider the cases for row s. If A s,vŷ 0, then by (1), we have (Aŷ) s = (Ay) s /q, implying (Ay) s < (Ay) i. On the other hand, if A s,vŷ = 0 but A s,s(y) 0, then C would not be a connected component of H (since a neighbor of s would be missing). Thus A s,s(y) = 0 and (Ay) s = (Aŷ) s = 0, and (Ay) i = q(aŷ) i > 0. In both cases we contradict that (x, y) is an equilibrium. Thus we fulfill condition (i) of Lemma 1, and by symmetry we also fulfill condition (ii). We have shown that (ˆx, ŷ) is an equilibrium, contradicting the minimality of (x, y). As an immediate corollary of Lemma 3 and 4, we get that the subgraph in G(G) induced by the extended support of a minimal equilibrium has at most two connected components. In the following lemma we show that in graphs of small maximum degree, we can find all such subgraphs quite efficiently. This will allow us to find a small, minimal equilibrium by checking all sets of rows and columns that would be candidates for being the extended supports of one. Lemma 5. Let G be a graph on n vertices and with maximum degree = (G). In time ( + 1) 2t n c+o(1) one can enumerate all subgraphs on t vertices that consist of c connected components. Proof. We first show how to enumerate all connected subgraphs on t vertices in time ( + 1) 2t n O(1) by a branching algorithm. At any point selected vertices will be active or passive. When a vertex is selected it will first be active and later be set to passive. Selecting a vertex and making it active respectively setting a vertex to passive is called an event. First, we branch on the choice of one out of n starting vertices in G and set it active. Then until we have selected t vertices we do the following: We consider the least recently added active vertex and branch on one of at most + 1 events, namely selecting one of its at most neighbors (and making it active) or setting the vertex itself to passive. We terminate when we have selected t vertices (and output the corresponding subgraph) or when there are no more active vertices. Clearly, on each branch of this algorithm there are at most 2t events. Thus the branching tree has at most ( + 1) 2t leaves, implying a total runtime of ( +1) 2t n O(1). Observe that for every connected subgraph on t vertices there is a sequence of events such that the graph occurs in the enumeration. 7
8 8 The generalization to c components is straightforward: When there are no more active vertices but we have not yet selected c components, then we select one of the less than n remaining vertices of G as a new active vertex, starting a new component. Selecting new starting vertices adds a factor of n c to our runtime. We are now in position to describe our entire algorithm. It first iterates through all possible sizes of extended support in increasing order. In each iteration, it enumerates all subgraphs that might correspond to the extended support of a minimal equilibrium. It then checks all ways of selecting a support from the given subgraph, and for each such selection it uses the algorithm behind Theorem 1 to check whether there is an equilibrium on the support. If no equilibrium is found throughout the whole process, the algorithm reports that there exists no equilibrium with support size at most k in G. The running time is bounded by l O(kl) n O(1) from Lemma 5, times ( ) kl+k 2 k = 2 O(kl) ways of selecting the support, times n O(1) for checking for an equilibrium. In total, we get a running time of l O(kl) n O(1). Finally, completeness comes from the exhaustiveness of Lemma 5 and the structure given by Lemmas 3 and Non-negative payoffs In the case that the payoffs of our games are non-negative, i.e., A, B Q n n 0, we can reduce our running time to be polynomial in l, for l-sparse games. We begin with a strengthening of Lemmas 3 and 4. Lemma 6. Let G = (A, B) be a bimatrix game with A, B Q n n 0, and G be the graph associated with G. If (x, y) is a minimal Nash equilibrium for G, then either S(x) = S(y) = 1, or G[S(x) S(y)] is connected. Proof. Let G S := G[S(x) S(y)]; assume that G S is not connected. If the expected outcome is zero, then (since the entries are non-negative) every entry in A S(x), and B,S(y) is zero, and we get an equilibrium by selecting any single row i S(x) and column j S(y). Otherwise, every row of A S(x),S(y) and every column of B S(x),S(y) contains some positive entry. Let C be a connected component of G S on row vertices Vˆx and column vertices Vŷ, and define a new pair of strategy profiles (ˆx, ŷ) where S(ˆx) = Vˆx and S(ŷ) = Vŷ, by normalizing (x, y) onto Vˆx and Vŷ as in Lemma 4. We will argue that (ˆx, ŷ) is an equilibrium. Let s Vˆx, and assume by way of contradiction that for some row i [n], we have (Aŷ) i > (Aŷ) s. Let (Aŷ) s = c 0 ; by non-connectivity of G S, (Ay) s = qc 0. Further let (Aŷ) i = c 1 and (Ay) i = qc 1 + (1 q)c 2. Now (Ay) s = qc 0 < qc 1 qc 1 + (1 q)c 2 = (Ay) i, contradicting our assumptions; the last inequality is because the entries are non-negative. Repeating the argument symmetrically, we find that (ˆx, ŷ) is a Nash equilibrium. Thus, to find an equilibrium in G = (A, B), it suffices to search for occurrences of the support, rather than the extended support. Invoking Lemma 5 directly with a bound of 2k vertices gives a running time of l O(k) n O(1).
9 9 3.2 No polynomial kernels We next show another interesting application of the connectivity lemmas described in the section above. In particular, we will use Lemma 6 to rule out the possibility of a polynomial kernel for the problem of determining whether a Nash equilibrium with support sizes at most k exists in an l-sparse bimatrix game (see e.g. [3] for a discussion on the important concept of polynomial-kernelization). In our setting, a polynomial kernel is an algorithm that receives as input an l-sparse game G and an integer k, and outputs in polynomial-time an l -sparse game G and an integer k, with G + k + l p(l, k ) for some polynomial p, such that G has an equilibrium with support sizes bounds by k iff G has an equilibrium with support sizes bounds by k. A framework for providing evidence that polynomial kernels do not exist for a given problem was given in [3]. There, the concept of composition algorithms play a central role. A composition algorithm for a parameterized problem Π is an algorithm that receives as input a sequence of instances (x 1, k),..., (x t, k) of Π, all sharing the same parameter, and output in time polynomial in i x i an instance (y, k ) of Π such that (y, k ) Π i : (x i, k) Π and k p(k) for some polynomial p. A parameterized problem admitting a composition algorithm is called compositional. The following connection between compositional problems and polynomial kernels was proven in [3] and [21]: Theorem 6 ([3, 21]). Unless co-np NP/poly, no compositional parameterized problem whose unparameterized variant is NP-complete has a polynomial kernel. Observe that the unparameterized decision variant of our problem is the problem of determining whether there exists an equilibrium in a given bimatrix game with support sizes bounded by k (since l is part of the input, the sparseness does not come into affect here). This problem was shown to be NP-complete by [22]. Thus, in order to show the non-existence of a polynomial kernel in our setting, we need to show that our problem is compositional. We will restrict ourselves to instances with non-negative payoffs. Given a sequence of l-sparse games (A 1, B 1 ),..., (A t, B t ) with non-negative payoffs, and a parameter k, our algorithm outputs the game G := (A, B) defined by A B A B 2 0 A := and B := A t 0 0 B t It is clear that our algorithm runs in polynomial-time, and that G is l-sparse. Furthermore, it is not difficult to see that due to Lemma 6, we know that G has an equilibrium with support sizes bounded by k iff (A i, B i ) has an equilibrium with support sizes bounded by k, for some i {1,..., t}. It follows that the above algorithm is a composition algorithm, and so according to Theorem 6, we can rule out the likelihood of polynomial kernels for our problem.
10 10 4 Locally Bounded Treewidth Games Let G = (A, B) be a given game with A, B P n n, with P Q, P l, and let G = G(G) the graph associated with G. In this section we will present an algorithm that finds an equilibrium with support sizes at most k when G comes from a graph class with locally bounded treewidth. Note that this is a partial extension of the results of the previous section, as graphs of bounded degree have locally bounded treewidth, while on the other hand we assume that there is a bounded set P of only l different payoff values which can occur in the games. (The case P = {0, 1} would correspond to win-lose games.) Definition 3 ([18]). A graph class has locally bounded treewidth if there is a function f : N N such that for every graph G := (V, E) of the class, any vertex v V, and any d N, the subgraph of G induced by all vertices within distance at most d from v has treewidth at most f(d). We refer readers to [20] for more details on the notion of treewidth and locally bounded treewidth. The crucial property of locally bounded treewidth graphs in our context is that first-order queries can be answered in FPT time on such graphs when the parameter is the size of first-order formula [20, Chapter 12.2]. For ease of presentation we show how to find an equilibrium where both players have support size k (the algorithm can be easily adapted to support sizes k 1, k 2 k). Let I and J be two subsets of k elements in [n]. We say that two matrices A, B Q k k occur in G at (I, J) if A = A I,J and B = B I,J. The pair (A, B ) forms an equilibrium pattern if there exists an equilibrium (x, y) where (A, B ) occur in G at (S(x), S(y)). Our algorithm will try all possible l 2k2 pairs of matrices (A, B ), and for each such pair it will determine whether it is an equilibrium pattern. When does a pair of matrices (A, B ) form an equilibrium pattern? The first obvious condition is that it occurs in G at some pair of position sets (I, J). Furthermore, by definition of a Nash equilibrium, there is a pair of strategies (x, y) with S(x) = I and S(y) = J, such that neither player has a better alternative. The difficulty here lies in the fact that, even given the support S(y) of the column player, there may be too many possible strategies for the row player that have supports different from I. To circumvent this, we define equivalence of rows with respect to supports S(y), and of columns with respect to supports S(x). Definition 4. Let I, J [n]. Two rows i 1, i 2 [n] are J-equivalent if A i1,j = A i2,j. Similarly, two columns j 1, j 2 [n] are I-equivalent if B I,j1 = B I,j2. Lemma 7. Let J be the support of the column player. For any row strategy x there is a row strategy ˆx such that: (i) the support S(ˆx) contains at most one row from each J-equivalence class (ii) and for any column strategy y with support J we have ˆx T Ay = x T Ay. The same is true for column strategies, given a support I of the row player.
11 11 Proof. We prove the lemma row strategies given a support J of the column player. The proof for column strategies is similar. Let x be a row strategy which includes two strategies i 1 and i 2 that are J-equivalent. It suffices to show that there is a row strategy ˆx with S(ˆx) = S(x) \ i 2 and ˆx T Ay = x T Ay for any column strategy y with support J. For this, take ˆx to be the strategy defined by ˆx i1 := x i1 + x i2, ˆx i2 := 0, and ˆx i = x i for all i {1,..., n} \ {i 1, i 2 }. Let y be any column strategy with S(y) = J. By definition of J-equivalence, we have (Ay) i1 = (Ay) i2. Thus, we get that (ˆx T Ay) i1 + (ˆx T Ay) i2 = (x T Ay) i1 + (x T Ay) i2. Furthermore, since ˆx and x equal on all entries i [n] \ {i 1, i 2 }, we know that (ˆx T Ay) i = (x T Ay) i, i i 1,i 2 i i 1,i 2 and thus (ˆx T Ay) = i i 1,i 2 (ˆx T Ay) i + (ˆx T Ay) i1 + (ˆx T Ay) i2 = i i 1,i 2 (x T Ay) i + (x T Ay) i1 + (x T Ay) i2 = (x T Ay). Therefore ˆx T Ay = x T Ay for all column strategies y with S(y) = J. For each possible equilibrium pattern (A, B ) we do the following. For each choice of rows A P 1 k that do not occur in A and each choice of columns B P k 1 that do not occur in B, we create two matrices C = ( A 0 A 0 ) ( B and D = B 0 0 We then check if there is an equilibrium (x, y) in the game (C, D) with S(x) = S(y) = [k] using linear programming. If there is such an equilibrium, then we proceed as follows to find an occurrence of (A, B ) that avoids the rows and columns which were not chosen. For this let F 1 be the rows which occur neither in A nor in A and let F 2 be the columns which occur neither in B nor in B. We say that F 1 and F 2 are forbidden for (A, B ). We note that given (A, B ), a set of rows F 1 P 1 k, and a set of columns F 2 P k 1, one can write a firstorder formula of size bounded by some function in k and P = l to determine whether (A, B ) has an occurrence which avoids F 1 and F 2. Example 1. Consider a win-lose game (A, B) encoded into relations A and B such that A(r, c) is true iff A r,c = 1, and likewise for B. Then a 2 2 equilibrium where the pattern is two identity matrices, and the all one vector is a forbidden row and column, can be found with the formula ). r 1, r 2, c 1, c 2 A(r 1, c 1 ) A(r 1, c 2 ) A(r 2, c 1 ) A(r 2, c 2 ) B(r 1, c 1 ) B(r 1, c 2 ) B(r 2, c 1 ) B(r 2, c 2 ) r ( A(r, c 1 ) A(r, c 2 )) c ( B(r 1, c ) B(r 2, c )).
12 12 In general, with l different values, there would be l 1 relations A i and B i encoding the game, where A i (r, c) is true if A r,c = z i, for every z i P except the zero value. Since the number of possible choices for F 1 and F 2 is bounded by some function in k and l, and for each such choice we can determine whether F 1 and F 2 is a forbidden pair for (A, B ) in polynomial-time, the total time for determining whether (A, B ) is an equilibrium pattern is FPT in k and l. Since the number of pairs (A, B ) is also bounded by a function in k, the total running time of our entire algorithm is also FPT in k and l. To complete the proof of Theorem 5, let us briefly argue completeness. Assume that there is any equilibrium with support sizes equal to k, let I and J be the supports, and let A and B be corresponding sub-matrices. Observe that we may set all entries in columns outside J of A to zero without harm, ditto for rows outside I in B. According to Lemma 7 it suffices to keep one copy of each row outside A in A (also discard the corresponding zero-row in B to keep the size the same). The same is of course true for columns outside B in B. Except for a permutation this is equal to one of the games (C, D) that we considered. Therefore our algorithm will find such an equilibrium if one exists. 5 Unbalanced Games In this section we briefly consider k-unbalanced bimatrix games. A bimatrix game (A, B) is k-unbalanced if A, B Q k n 0 for some k << n [24, 25] (i.e., the row player has a significantly smaller number of strategies than the column player). We will show that a Nash equilibrium can be computed in FPT-time with respect to k and l, where l denotes the number of different payoffs that the row player has, i.e., l := {A i,j : 1 i k, 1 j n}. Similar to Definition 4 we define two column strategies i, j [n] to be equivalent if A,i = A,j. (However, notice that unlike Definition 4, here equivalence of column strategies is defined with respect to the row player payoff.) Lemma 8. For each equilibrium there is an equilibrium where the column player plays at most one column from each equivalence class. Proof. Let (x, y) be an equilibrium, and suppose that S(y) includes two equivalent strategies i and j. To prove the lemma it suffices to show that the there exists an equilibrium (x, ŷ) with S(ŷ) = S(y) \ {j}. For this, let us take ŷ to be the strategy vector defined by ŷ i := y i + y j, ŷ j := 0, and ŷ x = y x for all x {1,..., n} \ {i, j}. Clearly S(ŷ) = S(y) \ {j}. Thus, for each s S(ŷ) we have (x T B) s (x T B) j, j s, since this condition holds for all s S(y) S(ŷ) by Lemma 1. Furthermore, due to the definition of equivalence, a simple calculation shows that (Ay) s = (Aŷ) s for all s {1,..., k}. Therefore, for each s S(x), we get again using Lemma 1 that (Aŷ) s = (Ay) s (Ay) j, j s.
13 13 It follows that (x, ŷ) satisfies both conditions of Lemma 1, and so it is indeed an equilibrium. Using Lemma 8 we can easily devise an FPT algorithm for computing a Nash equilibrium in our setting. The algorithm simply guesses the support of the row player and column player, and then uses the method of Theorem 1 to determine whether there exists a Nash equilibrium corresponding to these sets of supports. Observe that there are at most l k column-strategy equivalence classes. Furthermore, according to Lipton, Markakis, and Mehta [25], in a k-unbalanced game there always exists an equilibrium where the column player has support size at most k + 1. Thus the number of guesses the algorithm makes is bounded by 2 k ( l k+1) k = l O(k 2), and for each such guess, the amount of time required is polynomial. This completes the proof of Theorem 4. 6 Conclusions This paper is among the first attempts at applying parameterized complexity techniques in algorithmic game theory. This seems surprising when considering the potential benefit both fields can enjoy from each other. Our paper focused on the fundamental game-theoretical problem of computing a Nash equilibrium in bimatrix game. Three parameterized algorithms were presented, each corresponding to a different parameterization of the problem. In two cases, our algorithms utilized the graph-theoretical structure inherited in bimatrix games, and we believe this perspective will be useful in other settings as well. Our work is only the first step towards completely understanding the multivariate complexity of computing Nash equilibria in bimatrix games. There are still several parameters of the problems which were left unexplored, and we consider the problem of identifying new parameterizations to be the central open problem of this paper. Below we list several other questions which would be interesting to explore: Can the running-time of Theorem 3 be substantially improved? Is there a polynomial-time algorithm for computing Nash equilibria in games whose bipartite graph representation has bounded treewidth? Can one remove the number of different values parameterizations from Theorem 4 and Theorem 5? If the answer to the previous question regarding Theorem 4 is no, how would negative evidence for this question look like? Note that the concept of PPADcompleteness does not immediately apply. References 1. T.G. Abbott, D.M. Kane, and P. Valiant. On the complexity of two-player win-lose games. In Proc. of the 46th annual IEEE symposium on Foundations Of Computer Science (FOCS), pages , 2005.
14 14 2. L. Addario-Berry, N. Olver, and A. Vetta. A polynomial time algorithm for finding Nash equilibria in planar win-lose games. Journal of Graph Algorithms and Applications, 11(1): , H.L. Bodlaender, R.G. Downey, M.R. Fellows, and D. Hermelin. On problems without polynomial kernels. Journal of Computer and System Sciences, 75(8): , H. Bosse, J. Byrka, and E. Markakis. New algorithms for approximate nash equilibria in bimatrix games. Theoretical Computer Science, 411(1): , J. Chen, B. Chor, M.R. Fellows, X. Huang, D.W. Juedes, I.A. Kanj, and G. Xia. Tight lower bounds for certain parameterized NP-hard problems. Information and Computation, 201(2): , X. Chen and X. Deng. 3-NASH is PPAD-complete. Electronic Colloquium on Computational Complexity, (134), X. Chen and X. Deng. Settling the complexity of two-player nash equilibrium. In Proc. of the 47th annual IEEE symposium on Foundations Of Computer Science (FOCS), pages , X. Chen, X. Deng, and S.-H. Teng. Computing nash equilibria: Approximation and smoothed complexity. In Proc. of the 47th annual IEEE symposium on Foundations Of Computer Science (FOCS), pages , X. Chen, X. Deng, and S.-H. Teng. Sparse games are hard. In Proc. of the 2nd international Workshop on Internet and Network Economics (WINE), pages , X. Chen, S.-H. Teng, and P. Valiant. The approximation complexity of win-lose games. In Proc. of the 18th annual ACM-SIAM Symposium On Discrete Algorithms (SODA), pages , B. Codenotti, M. Leoncini, and G. Resta. Efficient computation of nash equilibria for very sparse win-lose bimatrix games. In Proc. of the 14th Annual European Symposium on Algorithms (ESA), pages , C. Daskalakis, P.W. Goldberg, and C.H. Papadimitriou. The complexity of computing a nash equilibrium. In Proc. of the 38th annual ACM Symposium on Theory Of Computing (STOC), pages 71 78, C. Daskalakis, A. Mehta, and C.H. Papadimitriou. Progress in approximate nash equilibria. In Proc. of the 8th ACM Conference on Electronic Commerce (EC), pages , C. Daskalakis, A. Mehta, and C.H. Papadimitriou. A note on approximate nash equilibria. Theoretical Computer Science, 410(17): , C. Daskalakis and C.H. Papadimitriou. On oblivious ptas s for nash equilibrium. In Proc. of the 41st annual ACM Symposium on Theory Of Computing (STOC), pages 75 84, K. Daskalakis and C.H. Papadimitriou. Three-player games are hard. Electronic Colloquium on Computational Complexity, (139), R. Downey and M. Fellows. Parameterized Complexity. Springer-Verlag, D. Eppstein. Subgraph isomorphism in planar graphs and related problems. Journal of Graph Algorithms and Applications, 3(3), V. Estivill-Castro and M. Parsa. Computing nash equilibria gets harder: New results show hardness even for parameterized complexity. In Proc. of the 15th Computing: the Australasian Theory Symposium (CATS), volume 94, pages 81 87, J. Flum and M. Grohe. Parameterized Complexity Theory. Springer-Verlag, 2006.
15 21. L. Fortnow and R. Santhanam. Infeasibility of instance compression and succinct PCPs for NP. In Proc. of the 40th Annual ACM Symposium on Theory of Computing (STOC), pages , I. Gilboa and E. Zemel. Nash and correlated equillbrla: Some complexity considerations. Games and Economic Behavior, P.W. Goldberg and C.H. Papadimitriou. Reducibility among equilibrium problems. In Proc. of the 38th annual ACM Symposium on Theory Of Computing (STOC), pages 61 70, S. Kalyanaraman and C. Umans. Algorithms for playing games with limited randomness. pages , R.J. Lipton, E. Markakis, and A. Mehta. Playing large games using simple strategies. In Proc. of the 4th ACM Conference on Electronic Commerce (EC), pages 36 41, N. Nisan, T. Roughgarden, É. Tardos, and V.V. Vazirani. Algorithmic Game Theory. Cambridge University Press, C.H. Papadimitriou. Algorithms, games, and the internet. In Proc. of the 33rd Annual ACM Symposium on Theory of Computing (STOC), pages ,
Single parameter FPT-algorithms for non-trivial games
Single parameter FPT-algorithms for non-trivial games Author Estivill-Castro, Vladimir, Parsa, Mahdi Published 2011 Journal Title Lecture Notes in Computer science DOI https://doi.org/10.1007/978-3-642-19222-7_13
More informationTheoretical Computer Science
Theoretical Computer Science 0 (009) 599 606 Contents lists available at ScienceDirect Theoretical Computer Science journal homepage: www.elsevier.com/locate/tcs Polynomial algorithms for approximating
More informationKernelization Lower Bounds: A Brief History
Kernelization Lower Bounds: A Brief History G Philip Max Planck Institute for Informatics, Saarbrücken, Germany New Developments in Exact Algorithms and Lower Bounds. Pre-FSTTCS 2014 Workshop, IIT Delhi
More information1 PROBLEM DEFINITION. i=1 z i = 1 }.
Algorithms for Approximations of Nash Equilibrium (003; Lipton, Markakis, Mehta, 006; Kontogiannis, Panagopoulou, Spirakis, and 006; Daskalakis, Mehta, Papadimitriou) Spyros C. Kontogiannis University
More informationThe complexity of uniform Nash equilibria and related regular subgraph problems
The complexity of uniform Nash equilibria and related regular subgraph problems Vincenzo Bonifaci a,b,1,, Ugo Di Iorio b, Luigi Laura b a Dipartimento di Ingegneria Elettrica, Università dell Aquila. Monteluco
More informationApproximate Nash Equilibria with Near Optimal Social Welfare
Proceedings of the Twenty-Fourth International Joint Conference on Artificial Intelligence (IJCAI 015) Approximate Nash Equilibria with Near Optimal Social Welfare Artur Czumaj, Michail Fasoulakis, Marcin
More informationLarge Supports are required for Well-Supported Nash Equilibria
Large Supports are required for Well-Supported Nash Equilibria Yogesh Anbalagan 1, Hao Huang 2, Shachar Lovett 3, Sergey Norin 4, Adrian Vetta 5, and Hehui Wu 6 Abstract We prove that for any constant
More informationOn the hardness of losing width
On the hardness of losing width Marek Cygan 1, Daniel Lokshtanov 2, Marcin Pilipczuk 1, Micha l Pilipczuk 1, and Saket Saurabh 3 1 Institute of Informatics, University of Warsaw, Poland {cygan@,malcin@,mp248287@students}mimuwedupl
More informationA Note on Approximate Nash Equilibria
A Note on Approximate Nash Equilibria Constantinos Daskalakis, Aranyak Mehta, and Christos Papadimitriou University of California, Berkeley, USA. Supported by NSF grant CCF-05559 IBM Almaden Research Center,
More informationNew Algorithms for Approximate Nash Equilibria in Bimatrix Games
New Algorithms for Approximate Nash Equilibria in Bimatrix Games Hartwig Bosse Jaroslaw Byrka Evangelos Markakis Abstract We consider the problem of computing additively approximate Nash equilibria in
More informationLower bounds for polynomial kernelization
Lower bounds for polynomial kernelization Part 2 Micha l Pilipczuk Institutt for Informatikk, Universitetet i Bergen August 20 th, 2014 Micha l Pilipczuk Kernelization lower bounds, part 2 1/33 Outline
More informationOn the hardness of losing width
On the hardness of losing width Marek Cygan 1, Daniel Lokshtanov 2, Marcin Pilipczuk 1, Micha l Pilipczuk 1, and Saket Saurabh 3 1 Institute of Informatics, University of Warsaw, Poland {cygan@,malcin@,mp248287@students}mimuwedupl
More informationThe Complexity of the Permanent and Related Problems
The Complexity of the Permanent and Related Problems Tim Abbott and Alex Schwendner May 9, 2007 Contents 1 The Permanent 2 1.1 Variations on the Determinant...................... 2 1.2 Graph Interpretation...........................
More informationWell-Supported vs. Approximate Nash Equilibria: Query Complexity of Large Games
Well-Supported vs. Approximate Nash Equilibria: Query Complexity of Large Games Xi Chen 1, Yu Cheng 2, and Bo Tang 3 1 Columbia University, New York, USA xichen@cs.columbia.edu 2 University of Southern
More informationKernelization by matroids: Odd Cycle Transversal
Lecture 8 (10.05.2013) Scribe: Tomasz Kociumaka Lecturer: Marek Cygan Kernelization by matroids: Odd Cycle Transversal 1 Introduction The main aim of this lecture is to give a polynomial kernel for the
More informationPreprocessing of Min Ones Problems: A Dichotomy
Preprocessing of Min Ones Problems: A Dichotomy Stefan Kratsch and Magnus Wahlström Max-Planck-Institut für Informatik, Saarbrücken, Germany {skratsch,wahl}@mpi-inf.mpg.de Abstract. Min Ones Constraint
More informationApproximability and Parameterized Complexity of Consecutive Ones Submatrix Problems
Proc. 4th TAMC, 27 Approximability and Parameterized Complexity of Consecutive Ones Submatrix Problems Michael Dom, Jiong Guo, and Rolf Niedermeier Institut für Informatik, Friedrich-Schiller-Universität
More informationEfficient Nash Equilibrium Computation in Two Player Rank-1 G
Efficient Nash Equilibrium Computation in Two Player Rank-1 Games Dept. of CSE, IIT-Bombay China Theory Week 2012 August 17, 2012 Joint work with Bharat Adsul, Jugal Garg and Milind Sohoni Outline Games,
More informationLecture 9 : PPAD and the Complexity of Equilibrium Computation. 1 Complexity Class PPAD. 1.1 What does PPAD mean?
CS 599: Algorithmic Game Theory October 20, 2010 Lecture 9 : PPAD and the Complexity of Equilibrium Computation Prof. Xi Chen Scribes: Cheng Lu and Sasank Vijayan 1 Complexity Class PPAD 1.1 What does
More informationSettling Some Open Problems on 2-Player Symmetric Nash Equilibria
Settling Some Open Problems on 2-Player Symmetric Nash Equilibria Ruta Mehta Vijay V. Vazirani Sadra Yazdanbod College of Computing, Georgia Tech rmehta,vazirani,syazdanb@cc.gatech.edu Abstract. Over the
More informationTwo Edge Modification Problems Without Polynomial Kernels
Two Edge Modification Problems Without Polynomial Kernels Stefan Kratsch and Magnus Wahlström Max-Planck-Institut für Informatik, 66123 Saarbrücken, Germany Abstract. Given a graph G and an integer k,
More informationThe Complexity of Maximum. Matroid-Greedoid Intersection and. Weighted Greedoid Maximization
Department of Computer Science Series of Publications C Report C-2004-2 The Complexity of Maximum Matroid-Greedoid Intersection and Weighted Greedoid Maximization Taneli Mielikäinen Esko Ukkonen University
More informationAlgorithmic Game Theory and Economics: A very short introduction. Mysore Park Workshop August, 2012
Algorithmic Game Theory and Economics: A very short introduction Mysore Park Workshop August, 2012 PRELIMINARIES Game Rock Paper Scissors 0,0-1,1 1,-1 1,-1 0,0-1,1-1,1 1,-1 0,0 Strategies of Player A Strategies
More informationLecture December 2009 Fall 2009 Scribe: R. Ring In this lecture we will talk about
0368.4170: Cryptography and Game Theory Ran Canetti and Alon Rosen Lecture 7 02 December 2009 Fall 2009 Scribe: R. Ring In this lecture we will talk about Two-Player zero-sum games (min-max theorem) Mixed
More informationW -Hardness under Linear FPT-Reductions: Structural Properties and Further Applications
W -Hardness under Linear FPT-Reductions: Structural Properties and Further Applications Jianer Chen 1 Xiuzhen Huang 2 Iyad A. Kanj 3 Ge Xia 4 1 Dept. of Computer Science, Texas A&M University, College
More informationarxiv:cs/ v1 [cs.gt] 26 Feb 2006
On the Approximation and Smoothed Complexity of Leontief Market Equilibria arxiv:cs/0602090v1 [csgt] 26 Feb 2006 Li-Sha Huang Department of Computer Science Tsinghua University Beijing, China Abstract
More informationClassical Complexity and Fixed-Parameter Tractability of Simultaneous Consecutive Ones Submatrix & Editing Problems
Classical Complexity and Fixed-Parameter Tractability of Simultaneous Consecutive Ones Submatrix & Editing Problems Rani M. R, Mohith Jagalmohanan, R. Subashini Binary matrices having simultaneous consecutive
More informationDynamic Programming on Trees. Example: Independent Set on T = (V, E) rooted at r V.
Dynamic Programming on Trees Example: Independent Set on T = (V, E) rooted at r V. For v V let T v denote the subtree rooted at v. Let f + (v) be the size of a maximum independent set for T v that contains
More informationarxiv: v1 [cs.gt] 11 Feb 2011
On Oblivious PTAS s for Nash Equilibrium Constantinos Daskalakis EECS and CSAIL, MIT costis@mit.edu Christos Papadimitriou Computer Science, U.C. Berkeley christos@cs.berkeley.edu arxiv:1102.2280v1 [cs.gt]
More informationSome hard families of parameterised counting problems
Some hard families of parameterised counting problems Mark Jerrum and Kitty Meeks School of Mathematical Sciences, Queen Mary University of London {m.jerrum,k.meeks}@qmul.ac.uk September 2014 Abstract
More informationThe Complexity of Nash Equilibria as Revealed by Data
The Complexity of Nash Equilibria as Revealed by Data Siddharth Barman Umang Bhaskar Federico Echenique Adam Wierman Abstract In this paper we initiate the study of the computational complexity of Nash
More informationAn Improved Algorithm for Parameterized Edge Dominating Set Problem
An Improved Algorithm for Parameterized Edge Dominating Set Problem Ken Iwaide and Hiroshi Nagamochi Department of Applied Mathematics and Physics, Graduate School of Informatics, Kyoto University, Japan,
More informationCS364A: Algorithmic Game Theory Lecture #16: Best-Response Dynamics
CS364A: Algorithmic Game Theory Lecture #16: Best-Response Dynamics Tim Roughgarden November 13, 2013 1 Do Players Learn Equilibria? In this lecture we segue into the third part of the course, which studies
More informationA Cubic-Vertex Kernel for Flip Consensus Tree
To appear in Algorithmica A Cubic-Vertex Kernel for Flip Consensus Tree Christian Komusiewicz Johannes Uhlmann Received: date / Accepted: date Abstract Given a bipartite graph G = (V c, V t, E) and a nonnegative
More informationOn the Complexity of Two-Player Win-Lose Games
On the Complexity of Two-Player Win-Lose Games Tim Abbott, Daniel Kane, Paul Valiant April 7, 2005 Abstract The efficient computation of Nash equilibria is one of the most formidable computational-complexity
More informationFinding Consensus Strings With Small Length Difference Between Input and Solution Strings
Finding Consensus Strings With Small Length Difference Between Input and Solution Strings Markus L. Schmid Trier University, Fachbereich IV Abteilung Informatikwissenschaften, D-54286 Trier, Germany, MSchmid@uni-trier.de
More informationOn the Impossibility of Black-Box Truthfulness Without Priors
On the Impossibility of Black-Box Truthfulness Without Priors Nicole Immorlica Brendan Lucier Abstract We consider the problem of converting an arbitrary approximation algorithm for a singleparameter social
More informationParameterized Algorithms for Max Colorable Induced Subgraph problem on Perfect Graphs
Parameterized Algorithms for Max Colorable Induced Subgraph problem on Perfect Graphs Neeldhara Misra 1, Fahad Panolan 2, Ashutosh Rai 2, Venkatesh Raman 2, and Saket Saurabh 2 1 Indian Institute of Science,
More informationApproximating the minmax value of 3-player games within a constant is as hard as detecting planted cliques
Electronic Colloquium on Computational Complexity, Report No. 25 (2012) Approximating the minmax value of 3-player games within a constant is as hard as detecting planted cliques Kord Eickmeyer 1, Kristoffer
More informationCovering Games: Approximation through Non-Cooperation
Covering Games: Approximation through Non-Cooperation Martin Gairing Department of Computer Science, University of Liverpool, U.K. m.gairing@liverpool.ac.uk Abstract. We propose approximation algorithms
More information3 Nash is PPAD Complete
Electronic Colloquium on Computational Complexity, Report No. 134 (2005) 3 Nash is PPAD Complete Xi Chen Department of Computer Science Tsinghua University Beijing, P.R.China xichen00@mails.tsinghua.edu.cn
More informationprinceton univ. F 17 cos 521: Advanced Algorithm Design Lecture 17: Games, Min-Max and Equilibria, and PPAD
princeton univ. F 17 cos 521: Advanced Algorithm Design Lecture 17: Games, Min-Max and Equilibria, and PPAD Lecturer: Matt Weinberg Scribe:Sanjeev Arora Economic and game-theoretic reasoning specifically,
More informationA Note on Perfect Partial Elimination
A Note on Perfect Partial Elimination Matthijs Bomhoff, Walter Kern, and Georg Still University of Twente, Faculty of Electrical Engineering, Mathematics and Computer Science, P.O. Box 217, 7500 AE Enschede,
More informationBounded Budget Betweenness Centrality Game for Strategic Network Formations
Bounded Budget Betweenness Centrality Game for Strategic Network Formations Xiaohui Bei 1, Wei Chen 2, Shang-Hua Teng 3, Jialin Zhang 1, and Jiajie Zhu 4 1 Tsinghua University, {bxh08,zhanggl02}@mails.tsinghua.edu.cn
More informationarxiv: v1 [math.co] 28 Oct 2016
More on foxes arxiv:1610.09093v1 [math.co] 8 Oct 016 Matthias Kriesell Abstract Jens M. Schmidt An edge in a k-connected graph G is called k-contractible if the graph G/e obtained from G by contracting
More informationA Single-Exponential Fixed-Parameter Algorithm for Distance-Hereditary Vertex Deletion
A Single-Exponential Fixed-Parameter Algorithm for Distance-Hereditary Vertex Deletion Eduard Eiben a, Robert Ganian a, O-joung Kwon b a Algorithms and Complexity Group, TU Wien, Vienna, Austria b Logic
More informationA Fixed-Parameter Algorithm for Max Edge Domination
A Fixed-Parameter Algorithm for Max Edge Domination Tesshu Hanaka and Hirotaka Ono Department of Economic Engineering, Kyushu University, Fukuoka 812-8581, Japan ono@cscekyushu-uacjp Abstract In a graph,
More informationDiscrete Applied Mathematics
Discrete Applied Mathematics 159 (2011) 1345 1351 Contents lists available at ScienceDirect Discrete Applied Mathematics journal homepage: www.elsevier.com/locate/dam On disconnected cuts and separators
More informationA Note on the Complexity of Network Reliability Problems. Hans L. Bodlaender Thomas Wolle
A Note on the Complexity of Network Reliability Problems Hans L. Bodlaender Thomas Wolle institute of information and computing sciences, utrecht university technical report UU-CS-2004-001 www.cs.uu.nl
More informationFinding a Heaviest Triangle is not Harder than Matrix Multiplication
Finding a Heaviest Triangle is not Harder than Matrix Multiplication Artur Czumaj Department of Computer Science New Jersey Institute of Technology aczumaj@acm.org Andrzej Lingas Department of Computer
More informationThe minimum G c cut problem
The minimum G c cut problem Abstract In this paper we define and study the G c -cut problem. Given a complete undirected graph G = (V ; E) with V = n, edge weighted by w(v i, v j ) 0 and an undirected
More informationLimitations of Efficient Reducibility to the Kolmogorov Random Strings
Limitations of Efficient Reducibility to the Kolmogorov Random Strings John M. HITCHCOCK 1 Department of Computer Science, University of Wyoming Abstract. We show the following results for polynomial-time
More informationPolynomial kernels for constant-factor approximable problems
1 Polynomial kernels for constant-factor approximable problems Stefan Kratsch November 11, 2010 2 What do these problems have in common? Cluster Edge Deletion, Cluster Edge Editing, Edge Dominating Set,
More informationConvergence Rate of Best Response Dynamics in Scheduling Games with Conflicting Congestion Effects
Convergence Rate of est Response Dynamics in Scheduling Games with Conflicting Congestion Effects Michal Feldman Tami Tamir Abstract We study resource allocation games with conflicting congestion effects.
More informationThe Mixed Chinese Postman Problem Parameterized by Pathwidth and Treedepth
The Mixed Chinese Postman Problem Parameterized by Pathwidth and Treedepth Gregory Gutin, Mark Jones, and Magnus Wahlström Royal Holloway, University of London Egham, Surrey TW20 0EX, UK Abstract In the
More informationMINIMALLY NON-PFAFFIAN GRAPHS
MINIMALLY NON-PFAFFIAN GRAPHS SERGUEI NORINE AND ROBIN THOMAS Abstract. We consider the question of characterizing Pfaffian graphs. We exhibit an infinite family of non-pfaffian graphs minimal with respect
More informationParameterized Algorithms and Kernels for 3-Hitting Set with Parity Constraints
Parameterized Algorithms and Kernels for 3-Hitting Set with Parity Constraints Vikram Kamat 1 and Neeldhara Misra 2 1 University of Warsaw vkamat@mimuw.edu.pl 2 Indian Institute of Science, Bangalore neeldhara@csa.iisc.ernet.in
More informationOn disconnected cuts and separators
On disconnected cuts and separators Takehiro Ito 1, Marcin Kamiński 2, Daniël Paulusma 3 and Dimitrios M. Thilikos 4 1 Graduate School of Information Sciences, Tohoku University, Aoba-yama 6-6-05, Sendai,
More informationOn the Space Complexity of Parameterized Problems
On the Space Complexity of Parameterized Problems Michael Elberfeld Christoph Stockhusen Till Tantau Institute for Theoretical Computer Science Universität zu Lübeck D-23538 Lübeck, Germany {elberfeld,stockhus,tantau}@tcs.uni-luebeck.de
More informationarxiv: v1 [cs.ds] 2 Oct 2018
Contracting to a Longest Path in H-Free Graphs Walter Kern 1 and Daniël Paulusma 2 1 Department of Applied Mathematics, University of Twente, The Netherlands w.kern@twente.nl 2 Department of Computer Science,
More informationarxiv: v2 [cs.dm] 12 Jul 2014
Interval Scheduling and Colorful Independent Sets arxiv:1402.0851v2 [cs.dm] 12 Jul 2014 René van Bevern 1, Matthias Mnich 2, Rolf Niedermeier 1, and Mathias Weller 3 1 Institut für Softwaretechnik und
More informationCS364A: Algorithmic Game Theory Lecture #13: Potential Games; A Hierarchy of Equilibria
CS364A: Algorithmic Game Theory Lecture #13: Potential Games; A Hierarchy of Equilibria Tim Roughgarden November 4, 2013 Last lecture we proved that every pure Nash equilibrium of an atomic selfish routing
More informationA Separator Theorem for Graphs with an Excluded Minor and its Applications
A Separator Theorem for Graphs with an Excluded Minor and its Applications Noga Alon IBM Almaden Research Center, San Jose, CA 95120,USA and Sackler Faculty of Exact Sciences, Tel Aviv University, Tel
More informationSymmetries and the Complexity of Pure Nash Equilibrium
Symmetries and the Complexity of Pure Nash Equilibrium Felix Brandt a Felix Fischer a, Markus Holzer b a Institut für Informatik, Universität München, Oettingenstr. 67, 80538 München, Germany b Institut
More informationTight Approximation Ratio of a General Greedy Splitting Algorithm for the Minimum k-way Cut Problem
Algorithmica (2011 59: 510 520 DOI 10.1007/s00453-009-9316-1 Tight Approximation Ratio of a General Greedy Splitting Algorithm for the Minimum k-way Cut Problem Mingyu Xiao Leizhen Cai Andrew Chi-Chih
More informationThe Maximum Flow Problem with Disjunctive Constraints
The Maximum Flow Problem with Disjunctive Constraints Ulrich Pferschy Joachim Schauer Abstract We study the maximum flow problem subject to binary disjunctive constraints in a directed graph: A negative
More informationFirst Order Convergence and Roots
Article First Order Convergence and Roots Christofides, Demetres and Kral, Daniel Available at http://clok.uclan.ac.uk/17855/ Christofides, Demetres and Kral, Daniel (2016) First Order Convergence and
More informationNear-domination in graphs
Near-domination in graphs Bruce Reed Researcher, Projet COATI, INRIA and Laboratoire I3S, CNRS France, and Visiting Researcher, IMPA, Brazil Alex Scott Mathematical Institute, University of Oxford, Oxford
More informationExact and Approximate Equilibria for Optimal Group Network Formation
Exact and Approximate Equilibria for Optimal Group Network Formation Elliot Anshelevich and Bugra Caskurlu Computer Science Department, RPI, 110 8th Street, Troy, NY 12180 {eanshel,caskub}@cs.rpi.edu Abstract.
More informationA Multiplayer Generalization of the MinMax Theorem
A Multiplayer Generalization of the MinMax Theorem Yang Cai Ozan Candogan Constantinos Daskalakis Christos Papadimitriou Abstract We show that in zero-sum polymatrix games, a multiplayer generalization
More informationGAMES OF FIXED RANK: A HIERARCHY OF BIMATRIX GAMES
GAMES OF FIXED RANK: A HIERARCHY OF BIMATRIX GAMES RAVI KANNAN AND THORSTEN THEOBALD Abstract. We propose and investigate a hierarchy of bimatrix games (A, B), whose (entry-wise) sum of the pay-off matrices
More informationInduced Subgraph Isomorphism on proper interval and bipartite permutation graphs
Induced Subgraph Isomorphism on proper interval and bipartite permutation graphs Pinar Heggernes Pim van t Hof Daniel Meister Yngve Villanger Abstract Given two graphs G and H as input, the Induced Subgraph
More informationAn improved approximation algorithm for the stable marriage problem with one-sided ties
Noname manuscript No. (will be inserted by the editor) An improved approximation algorithm for the stable marriage problem with one-sided ties Chien-Chung Huang Telikepalli Kavitha Received: date / Accepted:
More informationSF2972 Game Theory Exam with Solutions March 15, 2013
SF2972 Game Theory Exam with s March 5, 203 Part A Classical Game Theory Jörgen Weibull and Mark Voorneveld. (a) What are N, S and u in the definition of a finite normal-form (or, equivalently, strategic-form)
More informationCombinatorial Agency of Threshold Functions
Combinatorial Agency of Threshold Functions Shaili Jain 1 and David C. Parkes 2 1 Yale University, New Haven, CT shaili.jain@yale.edu 2 Harvard University, Cambridge, MA parkes@eecs.harvard.edu Abstract.
More informationClaw-free Graphs. III. Sparse decomposition
Claw-free Graphs. III. Sparse decomposition Maria Chudnovsky 1 and Paul Seymour Princeton University, Princeton NJ 08544 October 14, 003; revised May 8, 004 1 This research was conducted while the author
More informationApproximating min-max (regret) versions of some polynomial problems
Approximating min-max (regret) versions of some polynomial problems Hassene Aissi, Cristina Bazgan, and Daniel Vanderpooten LAMSADE, Université Paris-Dauphine, France {aissi,bazgan,vdp}@lamsade.dauphine.fr
More informationComputability and Complexity Theory: An Introduction
Computability and Complexity Theory: An Introduction meena@imsc.res.in http://www.imsc.res.in/ meena IMI-IISc, 20 July 2006 p. 1 Understanding Computation Kinds of questions we seek answers to: Is a given
More informationMore on NP and Reductions
Indian Institute of Information Technology Design and Manufacturing, Kancheepuram Chennai 600 127, India An Autonomous Institute under MHRD, Govt of India http://www.iiitdm.ac.in COM 501 Advanced Data
More informationPolynomial-time Computation of Exact Correlated Equilibrium in Compact Games
Polynomial-time Computation of Exact Correlated Equilibrium in Compact Games ABSTRACT Albert Xin Jiang Department of Computer Science University of British Columbia Vancouver, BC, Canada jiang@cs.ubc.ca
More informationCleaning Interval Graphs
Cleaning Interval Graphs Dániel Marx and Ildikó Schlotter Department of Computer Science and Information Theory, Budapest University of Technology and Economics, H-1521 Budapest, Hungary. {dmarx,ildi}@cs.bme.hu
More informationApproximate Well-supported Nash Equilibria Below Two-thirds
Algorithmica (06) 76:97 9 DOI 0.007/s0045-05-009- Approximate Well-supported Nash Equilibria Below Two-thirds John Fearnley Paul W. Goldberg Rahul Savani Troels Bjerre Sørensen Received: December 04 /
More information6.891 Games, Decision, and Computation February 5, Lecture 2
6.891 Games, Decision, and Computation February 5, 2015 Lecture 2 Lecturer: Constantinos Daskalakis Scribe: Constantinos Daskalakis We formally define games and the solution concepts overviewed in Lecture
More informationGame Theory. Greg Plaxton Theory in Programming Practice, Spring 2004 Department of Computer Science University of Texas at Austin
Game Theory Greg Plaxton Theory in Programming Practice, Spring 2004 Department of Computer Science University of Texas at Austin Bimatrix Games We are given two real m n matrices A = (a ij ), B = (b ij
More informationFixed Parameter Algorithms for Interval Vertex Deletion and Interval Completion Problems
Fixed Parameter Algorithms for Interval Vertex Deletion and Interval Completion Problems Arash Rafiey Department of Informatics, University of Bergen, Norway arash.rafiey@ii.uib.no Abstract We consider
More informationA Note on the Approximation of Mean-Payoff Games
A Note on the Approximation of Mean-Payoff Games Raffaella Gentilini 1 1 University of Perugia, Italy Abstract. We consider the problem of designing approximation schemes for the values of mean-payoff
More informationarxiv: v1 [cs.gt] 31 Mar 2019
On the Computation of Strategically Equivalent Rank-0 Games Joseph L Heyman Abhishek Gupta arxiv:190400450v1 [csgt] 31 Mar 2019 April 2, 2019 Abstract It has been well established that in a bimatrix game,
More informationarxiv: v1 [cs.gt] 4 Apr 2017
Communication Complexity of Correlated Equilibrium in Two-Player Games Anat Ganor Karthik C. S. Abstract arxiv:1704.01104v1 [cs.gt] 4 Apr 2017 We show a communication complexity lower bound for finding
More informationA New Upper Bound for Max-2-SAT: A Graph-Theoretic Approach
A New Upper Bound for Max-2-SAT: A Graph-Theoretic Approach Daniel Raible & Henning Fernau University of Trier, FB 4 Abteilung Informatik, 54286 Trier, Germany {raible,fernau}@informatik.uni-trier.de Abstract.
More informationComplexity Theory VU , SS The Polynomial Hierarchy. Reinhard Pichler
Complexity Theory Complexity Theory VU 181.142, SS 2018 6. The Polynomial Hierarchy Reinhard Pichler Institut für Informationssysteme Arbeitsbereich DBAI Technische Universität Wien 15 May, 2018 Reinhard
More informationOutline. Complexity Theory EXACT TSP. The Class DP. Definition. Problem EXACT TSP. Complexity of EXACT TSP. Proposition VU 181.
Complexity Theory Complexity Theory Outline Complexity Theory VU 181.142, SS 2018 6. The Polynomial Hierarchy Reinhard Pichler Institut für Informationssysteme Arbeitsbereich DBAI Technische Universität
More informationA NOTE ON STRATEGY ELIMINATION IN BIMATRIX GAMES
A NOTE ON STRATEGY ELIMINATION IN BIMATRIX GAMES Donald E. KNUTH Department of Computer Science. Standford University. Stanford2 CA 94305. USA Christos H. PAPADIMITRIOU Department of Computer Scrence and
More informationarxiv:quant-ph/ v1 15 Apr 2005
Quantum walks on directed graphs Ashley Montanaro arxiv:quant-ph/0504116v1 15 Apr 2005 February 1, 2008 Abstract We consider the definition of quantum walks on directed graphs. Call a directed graph reversible
More informationGraph coloring, perfect graphs
Lecture 5 (05.04.2013) Graph coloring, perfect graphs Scribe: Tomasz Kociumaka Lecturer: Marcin Pilipczuk 1 Introduction to graph coloring Definition 1. Let G be a simple undirected graph and k a positive
More informationFrom Primal-Dual to Cost Shares and Back: A Stronger LP Relaxation for the Steiner Forest Problem
From Primal-Dual to Cost Shares and Back: A Stronger LP Relaxation for the Steiner Forest Problem Jochen Könemann 1, Stefano Leonardi 2, Guido Schäfer 2, and Stefan van Zwam 3 1 Department of Combinatorics
More informationA New Lower Bound on the Maximum Number of Satisfied Clauses in Max-SAT and its Algorithmic Applications
A New Lower Bound on the Maximum Number of Satisfied Clauses in Max-SAT and its Algorithmic Applications Robert Crowston, Gregory Gutin, Mark Jones, Anders Yeo Department of Computer Science Royal Holloway,
More informationof a bimatrix game David Avis McGill University Gabriel Rosenberg Yale University Rahul Savani University of Warwick
Finding all Nash equilibria of a bimatrix game David Avis McGill University Gabriel Rosenberg Yale University Rahul Savani University of Warwick Bernhard von Stengel London School of Economics Nash equilibria
More informationChain Minors are FPT. B Marcin Kamiński. 1 Introduction. Jarosław Błasiok 1 Marcin Kamiński 1
Algorithmica (2017) 79:698 707 DOI 10.1007/s00453-016-0220-1 Chain Minors are FPT Jarosław Błasiok 1 Marcin Kamiński 1 Received: 21 March 2014 / Accepted: 21 September 2016 / Published online: 20 October
More informationk-distinct In- and Out-Branchings in Digraphs
k-distinct In- and Out-Branchings in Digraphs Gregory Gutin 1, Felix Reidl 2, and Magnus Wahlström 1 arxiv:1612.03607v2 [cs.ds] 21 Apr 2017 1 Royal Holloway, University of London, UK 2 North Carolina State
More informationDetecting Backdoor Sets with Respect to Horn and Binary Clauses
Detecting Backdoor Sets with Respect to Horn and Binary Clauses Naomi Nishimura 1,, Prabhakar Ragde 1,, and Stefan Szeider 2, 1 School of Computer Science, University of Waterloo, Waterloo, Ontario, N2L
More information